By Robbin Laird
A useful analytical distinction has been drawn by Paul Bracken between how alliances are understood in political science and how they function in game theory. In political science, alliances are negotiated documents — formal texts that commit states to one another across a range of contingencies. In game theory, alliances do not exist as general artifacts at all; they resolve into a series of issue-specific calculations about whether, in any given scenario, a state’s incentive structure makes follow-through rational.
This is not merely an academic distinction. It maps directly onto some of the most consequential strategic questions of the current era: whether American commitments in Asia and Europe will hold under pressure, whether adversaries will test those commitments, and whether the architecture of deterrence is as robust as the treaty texts suggest.
The Political Science View: Alliances as Negotiated Documents
In the mainstream political science and international relations tradition, an alliance is above all a text. It is a formal agreement negotiated between sovereign states, committing each party to specified obligations toward the others. The North Atlantic Treaty’s Article 5, the ANZUS Treaty, the Mutual Defense Treaty with Japan, the Taiwan Relations Act, these are the touchstones of American alliance architecture, and they are understood primarily as documents whose credibility depends on whether the signatory will honor its written terms.
This framework has a number of important features.
First, the unit of analysis is the bilateral or multilateral relationship as a whole, the alliance is a general commitment that covers a range of contingencies under a single umbrella.
Second, credibility is understood in reputational and systemic terms: a state that fails to honor its alliance commitments pays a cost in prestige and reliability that ripples across all of its relationships, not just the specific one where the failure occurred.
This is the logic that made American credibility in Vietnam seem important to American credibility in Europe, the reasoning that alliance reliability is indivisible.
Third, alliances in this framework accumulate institutional weight over time. NATO is not merely a treaty; it is a headquarters, a command structure, decades of joint exercises, deep interoperability between forces, and a political community with its own internal legitimacy. That institutional weight makes the commitment more credible and exit more costly.
The political science framework is also the natural home of alliance management, the diplomatic work of maintaining the text, renewing commitments, negotiating burden-sharing, and managing the gap between what the treaty says and what the partners are actually capable of delivering. It is a framework suited to foreign ministries, to summit meetings, and to the declaratory layer of strategy.
The weakness of this framework is that it treats the document as the operative reality.
But as any practitioner knows — and as decades of strategic studies have confirmed — what states actually do in a crisis may bear little resemblance to what the treaty text requires.
The gap between commitment and capability, between declared policy and actual behavior, between the document and the decision is precisely where deterrence fails.
The Game Theory View: Alliances as Issue-Specific Equilibria
Game theory approaches alliances from an entirely different direction. Rather than asking whether a commitment exists in documentary form, it asks whether follow-through is rational in a given scenario. Alliances, in this framework, are not general artifacts. They are equilibrium outcomes in repeated strategic interactions, and their meaning is always issue-specific.
The foundational insight here runs from Schelling’s work on commitment and coercion through the formal deterrence literature of the 1980s and 1990s and into contemporary work on credibility and signaling. The key question is not whether a treaty exists but whether the patron has sufficient incentive to act at the moment of crisis, whether its payoffs in the specific game being played make intervention more attractive than non-intervention.
A treaty, in this framework, is at best a coordination device or a costly signal about underlying preferences. It does not solve the time-consistency problem: the question is always whether the commitment is credible at the moment of execution, not at the moment of signing.
This leads directly to the issue-specific character of alliances in game theory. Whether the United States would fight to defend Taiwan is a different game than whether it would fight to defend Estonia, which is different again from whether it would fight to defend the Philippines in the South China Sea.
Each scenario has a distinct payoff structure, a distinct escalation ladder, and a distinct set of domestic political constraints. The treaty text may be identical or similar across these cases but the game is not. An adversary that conducts a game-theoretic analysis does not simply ask whether an alliance exists; it asks whether, in this specific scenario, with these specific stakes, the ally’s incentives support follow-through.
The game theory framework also illuminates the logic of salami tactics, the incremental probing that characterizes Chinese pressure on Taiwan, Russian behavior in the Baltic littoral, and Iranian activity in the Gulf.
Each slice of the salami is a separate game. By keeping each individual provocation below the threshold where intervention appears rational, an adversary can progressively shift the strategic landscape without triggering the alliance response that a single large aggression might provoke. The treaty text applies in principle to every slice but the payoff calculation that makes intervention rational applies to none of them individually.
The Analytical Gap and Its Strategic Consequences
The gap between these two frameworks is the central vulnerability of contemporary deterrence. Alliance management in the political science sense — maintaining the document, renewing commitments, managing burden-sharing — is necessary but not sufficient. What it cannot do is guarantee that in any specific scenario the ally’s payoff structure supports intervention. And it is precisely the payoff structure, not the treaty text, that adversaries read.
This generates a distinction between two kinds of strategic work that are often conflated in policy discussions.
The first is maintaining the document, the diplomatic and political work of keeping alliance texts current, institutions functioning, and rhetoric aligned.
The second is shaping the game, ensuring that in each specific contingency, the structure of payoffs makes allied follow-through rational, credible, and militarily feasible.
The first kind of work happens at foreign ministries and summits. The second happens in posture decisions, force deployments, pre-positioning, interoperability investments, rules of engagement, and the architecture of combined operations.
The challenge is that the second kind of work is less visible and politically less tractable than the first. A summit communiqué reaffirming alliance commitments is comprehensible to publics and politicians. The decision to pre-position forces in ways that make intervention automatic rather than discretionary, to create what strategists call tripwires or what contemporary distributed operations theorists call engagement architectures — is harder to explain and harder to sustain politically.
Yet it is precisely this second kind of work that addresses the game-theoretic vulnerability that the first kind leaves open.
The kill web concept that has emerged in American military thinking over the past decade is, among other things, an attempt to address this problem in operational terms. A distributed, networked force architecture in which sensors, shooters, and decision nodes are spread across the battlespace in which engagement is not sequential but simultaneous, not centralized but dispersed changes the payoff calculation in specific scenarios in ways that a treaty text cannot.
It makes intervention less a matter of political decision and more a matter of operational reality. The architecture itself is the commitment mechanism.
Disaggregation as an Adversary Strategy
Understanding the gap between the two frameworks also clarifies a key feature of contemporary adversary strategy: the deliberate effort to disaggregate alliances by manipulating the game-theoretic calculation on specific issues.
China, Russia, and Iran each pursue versions of this strategy, though with different instruments and in different theaters. The common logic is to identify scenarios in which the ally’s payoff calculation makes intervention irrational, and to construct crises within those scenarios rather than across the broader alliance relationship. This is why Chinese pressure on Taiwan is calibrated below the threshold of obvious aggression, why Russian probing in the gray zone short of Article 5 has been a recurring pattern, and why Iranian activity in the Strait of Hormuz and the Red Sea has been conducted through proxies and deniable means rather than through direct military action. Each of these approaches is designed to exploit the gap between what the treaty says and what the game makes rational.
The implication is that allied strategy must address not only the diplomatic maintenance of the document but the active shaping of payoff structures across the full range of scenarios that adversaries might exploit. This requires a level of strategic granularity, a scenario-by-scenario analysis of where the game-theoretic vulnerabilities lie, that general alliance management frameworks do not naturally provide. It requires asking not just whether the alliance is healthy in the aggregate but whether, in each specific contingency, the structure of incentives supports the outcome that deterrence requires.
Toward an Integrated Framework
The political science and game theory frameworks address different layers of the same strategic problem. The political science framework is right that alliances are more than moment-by-moment calculations; the institutional weight of NATO, the history of American treaty commitments, and the reputational costs of abandonment are real strategic facts that shape the payoff calculations adversaries make. The game theory framework is right that treaty texts do not automatically generate credible commitments; the operative question is always whether the specific scenario makes intervention rational.
What Bracken’s distinction points toward is the need for a two-level approach to alliance strategy. At the macro level, the work is diplomatic: maintaining the document, renewing commitments, managing the political health of the alliance relationship. At the micro level, the work is operational: shaping the specific game in each contingency, ensuring that payoff structures support allied intervention, and designing force architectures that make commitments credible not just declaratorily but mechanically.
The failure mode of contemporary Western alliance management is to concentrate effort at the macro level while allowing the micro-level vulnerabilities to accumulate. Summits are held, communiqués are issued, burden-sharing debates are conducted while in specific scenarios in the Taiwan Strait, the Baltic littoral, and the South China Sea, the game-theoretic calculation remains ambiguous. Adversaries understand this asymmetry.
Addressing it is the central challenge of alliance strategy in the current era.
The distinction between alliances as documents and alliances as games is not merely analytical. It is the lens through which the most consequential strategic questions of the current moment must be read.
Whether deterrence holds in the Indo-Pacific, whether NATO’s eastern flank remains secure, whether gray zone aggression can be contained, all of these depend not on the health of the document alone but on whether, in each specific scenario, the game has been shaped to make the right outcome rational.
That is the work that matters, and it is the work that is hardest to see.
Note: This is the first of three articles. The next deals with Pax Silica. And the third deals with Pax Silica as a case study of the analytical perspective of this article.